Bootslop: A Sourceless ColorForth Derivative

This repository contains the curation materials and prototype implementation for building a zero-overhead, sourceless ColorForth-derivative for x86-64, specifically modeled after the architectures of Timothy Lottes and Onat Türkçüoğlu.

Project Goal

The objective is to learn how to build this architecture from scratch, with the AI acting as a highly contextualized mentor.

Current State

The attempt_1/ directory contains a working C prototype that successfully implements the core architectural pillars:

• A "sourceless" editor that manipulates a 32-bit token array (Tape Drive) and a parallel 64-bit annotation array.
• A modal, interactive GUI built with raw Win32 GDI calls.
• A handmade Just-In-Time (JIT) compiler that translates tokens into executable x86-64 machine code on every keypress.
• An execution model based on Onat's 2-register stack (RAX/RDX) and a global memory tape.

Helper Scripts

This repository contains several Python scripts used during the initial curation and content-gathering phase:

• process_visuals.py: Downloads videos from YouTube, extracts frames based on transcript timestamps, performs OCR on the frames, and uses color analysis to generate semantically-tagged markdown logs of the visual content. It also crops out relevant code blocks and diagrams.
• fetch_blog.py: Parses TimothyLottesBlog.csv and scrapes the HTML content of each blog post, converting it to clean markdown for local archival.
• fetch_notes.py: Parses FORTH_NOTES.csv, filters out irrelevant or already-processed links, and scrapes the remaining pages into markdown files.
• estimate_context.py: A utility to scan the references/ directory and provide a rough estimate of the total token count to ensure it fits within the AI model's context window.
• ocr_interaction.py: A small utility to perform OCR on single image files.